This paper presents a model for partitioning two modes of three-way proximity data which generalizes INDCLUS by incorporating possible external information on objects and/or subjects. Specifically, subjects are partitioned into homogeneous classes, where class-conditional groups of objects are determined. The classifications of both objects and subjects are assumed to be related to possible external variables to better account for the meaning and the determinant of the groups. The model is fitted in a least-squares framework and an efficient ALS algorithm is given. An illustrative application to a benchmark data set is presented.
This paper presents a model for partitioning two modes of three-way proximity data which generalizes INDCLUS by incorporating possible external information on objects and/or subjects. Specifically, subjects are partitioned into homogeneous classes, where class-conditional groups of objects are determined. The classifications of both objects and subjects are assumed to be related to possible external variables to better account for the meaning and the determinant of the groups. The model is fitted in a least-squares framework and an efficient ALS algorithm is given. An illustrative application to a benchmark data set is presented.
A general model for INDCLUS with external information / Bocci, Laura; Vicari, Donatella. - STAMPA. - (2013), pp. 53-56. (Intervento presentato al convegno Cladag 2013 tenutosi a Modena, Italia nel 12-20 Settembre 2013).
A general model for INDCLUS with external information
BOCCI, Laura;VICARI, Donatella
2013
Abstract
This paper presents a model for partitioning two modes of three-way proximity data which generalizes INDCLUS by incorporating possible external information on objects and/or subjects. Specifically, subjects are partitioned into homogeneous classes, where class-conditional groups of objects are determined. The classifications of both objects and subjects are assumed to be related to possible external variables to better account for the meaning and the determinant of the groups. The model is fitted in a least-squares framework and an efficient ALS algorithm is given. An illustrative application to a benchmark data set is presented.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
